I feel like require-from-twitter really helps enforce the Unix philosophy. It's hard to write modules with multiple responsibilities in 140 characters.
Actually, there are several immutable FS-over-tinyURL projects. I have created some of them too. A real problem with Twitter is that you can get banned quickly for doing this.
Twitter allows deletions and explicitly requires you to process and honor deletion events in their TOS. They have even blocked API access to programs that don't honor deletions.
So, just as bad (or, maybe worse, in that you can't even keep around committed code that comes through this process, if the originating tweet was deleted) as npm.
Twitter API clients are required to delete all copies of tweet contents when a tweet is yanked, or Twitter will revoke API access, which might bring your production systems to a deserved halt.
You have left out a very crucial fact which completely destroys your argument: Twitter has no verification process that the tweet was deleted because you're NOT ADVERTISING THE TWEETS. You're using them as source code. So the public is NOT privy to what you're doing with the tweets! So Twitter would have to get a search warrant to determine that you aren't deleting tweets! Of course they don't care; they care about what you're revealing to the public.
Your computer has to download the tweet before it runs the tweet. It will be cached on your computer. Compliance with twitter TOS means you have to clear your cache when twitter says the tweet was deleted. Since you're using the tweet to run code, it doesn't hurt anything if you ignore the "deleted tweet" advisory.
This is also probably a snarky shot at npm [1], for those who lack context.
[1] https://news.ycombinator.com/item?id=11340510